AI

Protect AI lands a $13.5M investment to harden AI projects from attack

Comment

digital umbrella
Image Credits: Andrzej Wojcicki (opens in a new window) / Getty Images

Seeking to bring greater security to AI systems, Protect AI today raised $13.5 million in a seed-funding round co-led by Acrew Capital and Boldstart Ventures with participation from Knollwood Capital, Pelion Ventures and Aviso Ventures. Ian Swanson, the co-founder and CEO, said that the capital will be put toward product development and customer outreach as Protect AI emerges from stealth.

Protect AI claims to be one of the few security companies focused entirely on developing tools to defend AI systems and machine learning models from exploits. Its product suite aims to help developers identify and fix AI and machine learning security vulnerabilities at various stages of the machine learning life cycle, Swanson explains, including vulnerabilities that could expose sensitive data.

“As machine learning models usage grows exponentially in production use cases, we see AI builders needing products and solutions to make AI systems more secure, while recognizing the unique needs and threats surrounding machine learning code,” Swanson told TechCrunch in an email interview. “We have researched and uncovered unique exploits and provide tools to reduce risk inherent in [machine learning] pipelines.”

Swanson co-launched Protect AI with Daryan Dehghanpisheh and Badar Ahmed roughly a year ago. Swanson and Dehghanpisheh previously worked together at Amazon Web Services (AWS) on the AI and machine learning side of the business; Swanson was the worldwide leader at AWS’s AI customer solutions team and Dehghanpisheh was the global leader for machine learning solution architects. Ahmed became acquainted with Swanson while working at Swanson’s last startup, DataScience.com, which was acquired by Oracle in 2017. Ahmed and Swanson worked together at Oracle as well, where Swanson was the VP of AI and machine learning.

Protect AI’s first product, NB Defense, is designed to work within Jupyter Notebook, a digital notebook tool popular among data scientists within the AI community. (A 2018 GitHub analysis found that there were more than 2.5 million public Jupyter Notebooks in use at the time of the report’s publication, a number that’s almost certainly climbed since then.) NB Defense scans Jupyter notebooks for AI projects — which usually contain all the code, libraries and frameworks needed to train, run and test an AI system — for security risks and provides remediation suggestions.

What sort of problematic elements might an AI project notebook contain? Swanson suggests internal-use authentication tokens and other credentials, for one. NB Defense also looks for personally identifiable information (e.g., names and phone numbers) and open source code with a “nonpermissive” license that might prohibit it from being used in a commercial system.

Jupyter Notebooks are typically used as scratchpads rather than production environments, and most are locked safely away from prying eyes. According to an analysis by Dark Reading, fewer than 1% of the approximately 10,000 instances of Jupyter Notebook on the public web are configured for open access. But it’s true the exploits aren’t just theoretical. Last December, security firm Lightspin uncovered a method that could allow an attacker to run any code on a victim’s notebook across accounts on AWS SageMaker, Amazon’s fully managed machine learning service.

Other research firms, including Aqua Security, have found that improperly secured Jupyter Notebooks are vulnerable to Python-based ransomware and cryptocurrency mining attacks. In a 2020 Microsoft survey of businesses using AI, the majority said that they don’t have the right tools in place to secure their machine learning models.

It might be premature to sound the alarm bells. There’s no evidence that attacks are happening at scale, despite a Gartner report predicting an increase in AI cyberattacks through the end of this year. But Swanson makes the case that prevention is key.

“[Many] existing security code scanning solutions are not compatible with Jupyter notebooks. These vulnerabilities, and many more, are due to a lack of focus and innovation from current cybersecurity solution providers, and is the largest differentiation for Protect AI: Real threats and vulnerabilities that exist in AI systems, today,” Swanson said.

Beyond Jupyter Notebooks, Protect AI will work with common AI development tools, including Amazon SageMaker, Azure ML and Google Vertex AI Workbench, Swanson says. It’s available for free to start, with paid options to be introduced in the future.

“Machine learning is … complex and the pipelines delivering machine learning at scale create and multiply cybersecurity blind spots that evade current cybersecurity offerings, preventing important risks from being adequately understood and mitigated. Additionally, emerging compliance and regulatory frameworks continue to advance the need to harden AI systems’ data sources, models, and software supply chain to meet increased governance, risk management and compliance requirement,” Swanson continued. “Protect AI’s unique capabilities and deep expertise in the machine leaning lifecycle for enterprises and AI at scale helps enterprises of all sizes meet today’s and tomorrow’s unique, emerging and increasing requirements for a safer, more secure AI powered digital experience.”

That’s promising a lot. But Protect AI has the advantage of entering a market with relatively few direct competitors. Perhaps the closest is Resistant AI, which is developing AI systems to protect algorithms from automated attacks.

Protect AI, which is pre-revenue, isn’t revealing how many customers it has today. But Swanson claims that the company has secured “enterprises in the Fortune 500” across verticals, including finance, healthcare and life sciences, as well as energy, gaming, digital businesses and fintech.

“As we grow our customers, build partners and value chain participants we will use our funding to add additional team members in software development, engineering, security and go-to-market roles throughout 2023,” Swanson said, adding that Protect AI’s headcount stands at 15. “We have several years of cash runway available to continue to advance this field.”

More TechCrunch

Mobile app developers, including Patreon and Grammarly, are already integrating with Gemini Nano, its smallest AI model, the company announced during its I/O developer keynote on Tuesday. The companies, along…

Patreon and Grammarly are already experimenting with Gemini Nano, says Google

As part of the update, Reddit also launched a dedicated AMA tab within the web post composer.

Reddit introduces new tools for ‘Ask Me Anything,’ its Q&A feature

Here are quick hits of the biggest news from the keynote as they are announced.

Google I/O 2024: Here’s everything Google just announced

LearnLM is already powering features across Google products, including in YouTube, Google’s Gemini apps, Google Search and Google Classroom.

LearnLM is Google’s new family of AI models for education

The official launch comes almost a year after YouTube began experimenting with AI-generated quizzes on its mobile app. 

Google is bringing AI-generated quizzes to academic videos on YouTube

Around 550 employees across autonomous vehicle company Motional have been laid off, according to information taken from WARN notice filings and sources at the company.  Earlier this week, TechCrunch reported…

Motional cut about 550 employees, around 40%, in recent restructuring, sources say

The keynote kicks off at 10 a.m. PT on Tuesday and will offer glimpses into the latest versions of Android, Wear OS and Android TV.

Google I/O 2024: Watch all of the AI, Android reveals

It ran 110 minutes, but Google managed to reference AI a whopping 121 times during Google I/O 2024 (by its own count). CEO Sundar Pichai referenced the figure to wrap…

Google mentioned ‘AI’ 120+ times during its I/O keynote

Google Play has a new discovery feature for apps, new ways to acquire users, updates to Play Points, and other enhancements to developer-facing tools.

Google Play preps a new full-screen app discovery feature and adds more developer tools

Soon, Android users will be able to drag and drop AI-generated images directly into their Gmail, Google Messages and other apps.

Gemini on Android becomes more capable and works with Gmail, Messages, YouTube and more

Veo can capture different visual and cinematic styles, including shots of landscapes and timelapses, and make edits and adjustments to already-generated footage.

Google Veo, a serious swing at AI-generated video, debuts at Google I/O 2024

In addition to the body of the emails themselves, the feature will also be able to analyze attachments, like PDFs.

Gemini comes to Gmail to summarize, draft emails, and more

The summaries are created based on Gemini’s analysis of insights from Google Maps’ community of more than 300 million contributors.

Google is bringing Gemini capabilities to Google Maps Platform

Google says that over 100,000 developers already tried the service.

Project IDX, Google’s next-gen IDE, is now in open beta

The system effectively listens for “conversation patterns commonly associated with scams” in-real time. 

Google will use Gemini to detect scams during calls

The standard Gemma models were only available in 2 billion and 7 billion parameter versions, making this quite a step up.

Google announces Gemma 2, a 27B-parameter version of its open model, launching in June

This is a great example of a company using generative AI to open its software to more users.

Google TalkBack will use Gemini to describe images for blind people

Firebase Genkit is an open source framework that enables developers to quickly build AI into new and existing applications.

Google launches Firebase Genkit, a new open source framework for building AI-powered apps

This will enable developers to use the on-device model to power their own AI features.

Google is building its Gemini Nano AI model into Chrome on the desktop

Google’s Circle to Search feature will now be able to solve more complex problems across psychics and math word problems. 

Circle to Search is now a better homework helper

People can now search using a video they upload combined with a text query to get an AI overview of the answers they need.

Google experiments with using video to search, thanks to Gemini AI

A search results page based on generative AI as its ranking mechanism will have wide-reaching consequences for online publishers.

Google will soon start using GenAI to organize some search results pages

Google has built a custom Gemini model for search to combine real-time information, Google’s ranking, long context and multimodal features.

Google is adding more AI to its search results

At its Google I/O developer conference, Google on Tuesday announced the next generation of its Tensor Processing Units (TPU) AI chips.

Google’s next-gen TPUs promise a 4.7x performance boost

Google is upgrading Gemini, its AI-powered chatbot, with features aimed at making the experience more ambient and contextually useful.

Google’s Gemini updates: How Project Astra is powering some of I/O’s big reveals

Veo can generate few-seconds-long 1080p video clips given a text prompt.

Google’s image-generating AI gets an upgrade

At Google I/O, Google announced upgrades to Gemini 1.5 Pro, including a bigger context window. .

Google’s generative AI can now analyze hours of video

The AI upgrade will make finding the right content more intuitive and less of a manual search process.

Google Photos introduces an AI search feature, Ask Photos

Apple released new data about anti-fraud measures related to its operation of the iOS App Store on Tuesday morning, trumpeting a claim that it stopped over $7 billion in “potentially…

Apple touts stopping $1.8B in App Store fraud last year in latest pitch to developers

Online travel agency Expedia is testing an AI assistant that bolsters features like search, itinerary building, trip planning, and real-time travel updates.

Expedia starts testing AI-powered features for search and travel planning